I will write up an overview of the software next week if I can get onto it.
It is using deep learning, basically I am training the software with thousands of dirty/noisy images and clean images and then utilising the deep learning algorithms to clean up the shots.
There is a ton of prep at both ends before the algorithm kicks in, I’ll try and put together an explanation of how it all works that will make sense to people. I’m working on this on behalf of the company, so I won’t be able to share the code, but hopefully I can give enough detail to let others develop something similar themselves if it works well enough for production use.